Modern computer systems are frequently implemented as a distributed and/or virtualized system comprising a plurality of computer system services operating on a plurality of computer systems using a plurality of computer system resources. Such computer systems may allow connections from a large number of users and/or clients to allow access to the computer system services in order to perform a variety of tasks. Users and/or clients may connect to the computer system services using many different types of devices over a variety of different connection types. When a provider of computer system services wants to track and analyze attributes and/or metadata associated with the connection such as attributes and/or metadata related to the user, the device, the network and/or other such attributes and/or metadata, the provider can be faced with the problem of storing and processing very large data sets in a data store such as a data warehouse. In some computer system environments, where thousands of users may be performing tasks simultaneously, large data sets may be produced and stored in a data warehouse every second. A data warehouse, which may be statically configured in order to facilitate processing the large data sets, may require reconfiguring the data gathering service each time a new device, a new service, a new user type and/or other such new entities with new attributes and/or metadata may join the system. Additionally, a data warehouse may require reconfiguring the data gathering service each time the provider may wish to gather different data, or focus data gathering on a certain attribute subset or perform other types of analysis. Storage and manipulation of the large data sets as well as reconfiguring the data gathering service may be resource intensive, requiring expensive dedicated resources.